This model uses the Wio Terminal to recognize hand gestures: rock, paper and scissors using built-in Light sensor.
It is quite hard to accomplish using rule-based programming because gestures are not performed the same way all the time. Trying to solve this problem using traditional programming requires hundreds of different rules for each mode of action.
Even if we consider an idealized situation, we still miss a lot of motions such as speed, angle, or direction. A slight change in these factors requires a new set of rules to be defined but machine learning can handle these variations very easily.
A well-known possibility is to use camera sensors combined with machine learning to recognize gestures. Using a light sensor is equivalent to using only a one-pixel point of the camera to recognize gestures which is a completely different level of challenge.
Let's meet the challenge with this project.
More details on the project available on Hackster: https://www.hackster.io/vincekok/gesture-recognition-on-microcontroller-with-tinyml-72a124